Single-Channel Processing of Auscultatory Signals Using Methods of Mathematical Morphology

Abstract

The paper considers a new method of separating respiratory sounds from heart sounds in a general signal registered on the surface of the human body. The proposed approach is based on a combination of Bayesian noise suppression techniques and methods of mathematical morphology. The proposed method was tested on real auscultatory signals. Evaluation of the efficiency of the algorithm using auditory, visual and numerical analysis shows that the developed approach is a promising alternative to existing techniques for separating auscultatory signals into its natural components.